Microsoft Follows Mozilla In Considering Early Ban On SHA-1 Certificates (csoonline.com)

itwbennett writes: Following the first successful collision attack on the SHA-1 hashing algorithm last month, Mozilla said that it was considering a cut-off of July 1, 2016 to start rejecting all SHA-1 SSL certificates, ahead of an earlier scheduled date of Jan. 1, 2017. And now Microsoft is considering blocking the hashing algorithm on Windows by June next year.

Comments

Kickstarter

[nneonneo]

If it really is only $75-120K to crack SHA1, I propose we start a Kickstarter to gather the funds. Given the estimate of a few months, we'll ship our SHA1 collision well before a lot of other Kickstarter projects ship their products :)

Overrides

[sexconker]

At least let me fucking override shit for my devices (UPSes, copiers, etc.) that have absolutely no ability to use anything other than the self-signed shit they come with.

I'm fine with warning or blocking by default, but when those idiots remove my ability to do what I need to do (whitelist) I end up having to keep an older version of the browser with more holes in it just to connect to this UPS, that switch, this copier, etc.

Re:Overrides

[Zuriel]

You can join the ranks of people holding on to WinXP virtual machines because they need them to administer that one device that needs a certain version of Java 1.4 and Firefox 3.6.

Re:Overrides

[Gaygirlie]

Why not just use a portable version of some old Firefox, for example? If you use the portable, outdated version only for the outdated devices and the up-to-date Firefox everything else it shouldn't be too much of an issue?

Re:Overrides

[Big Hairy Ian]

The problem with the Internet of Things is that nobody seems to issue security updates for the plethora of devices we have plugged into our networks and when they do it's almost impossible to actually apply them.

Re: Overrides

[sexconker]

APC SmartUPS
Dell PowerConnect switches
HP switches (forget the name of the line)
Canon, Konica, and Xerox copiers
Etc.

These types of things are in place for decades, and all require some sort of ancient voodoo tool chain to fully manage.

Lord help you if you need to connect to an old UPS or switch via serial. 99 times out of 100 you won't have a serial port on any convenient device, the UPS to serial adapters only work half the time even when you can figure out the specific settings for the device, and then you run the risk running into a device that demands a non-standard cable with fucked-up pin mapping - I'm looking at you, APC.

It's gotten to the point that every device we deploy gets deployed with the serial cable permanently attached AND a USB adapter that is known to work with it. Then the device is covered in labels with shit like the internal IP, what flow control/etc. settings you need for serial, and which dark god you must pray to in order to get in and do your job.

Forced to click through

[Sits]

My experience of these changes is that you'll be forced to click through a warning in your browser even if you installed the certificate (or the root CA signing the certificate). The Microsoft page about no longer trusting SHA1 certs is confusing in this respect because it includes information about signing Windows binaries but it does say

Windows [...] will no longer trust any code that is signed with a SHA-1 code signing certificate and that contains a timestamp value greater than January 1, 2016

That document also says it only applies to certs that are in the Microsoft Root Certificate Program so ones you've manually installed might not be affected.

This is slightly different to the Mozilla's SHA-1 deprecation information:

After January 1, 2017, we plan to show the “Untrusted Connection” error whenever a SHA-1 certificate is encountered in Firefox.

Perhaps this isn't the override you were thinking of but it doesn't sound like a total block.

Re:git

[jonwil]

No, this only affects SSL certificates using the SHA-1 hash. Git isn't using the SHA-1 hash in a way where generating a collision would have security risks so there is no reason why anything has to change for Git.

Try not to be misguided

[GuB-42]

It's fine rejecting insecure certificates but sometimes, I'd rather have browsers get their priorities in order.
If you go on a SSL website that uses a self-signed certificate or use a slightly outdated one, you are presented with a scary warning page with multiple clicks needed to get to it. However, plain HTTP goes right through even though it is less secure than SSL with any bogus certificate.

Instead of a ban, I'm all for a rating system, like :
- Strong : everything OK, strong crypto
- Medium : slightly outdated, weaker crypto (SHA-1 could be on this level)
- Weak : self-signed, completely outdated
- None : HTTP
- Dangerous : revoked, mismatched certificate, suspect behavior (such as a decrease in security from last visit)
Only the "dangerous" category should trigger a warning, for the other categories, a different "lock" icon should be sufficient. Like the crossed-out "https" in Google Chrome.

Re:Try not to be misguided

[behrooz0az]

I quote GP:

suspect behavior (such as a decrease in security from last visit)

Re:Try not to be misguided

[GuB-42]

Indeed but posting sensitive data unencrypted is even worse and the browser won't say anything about it.
The problem is that the browser has no simple way of knowing if the site is sensitive or not. The best it can do is to tell you clearly about the level of security so that you can react accordingly.
"Dangerous" would be "worse that unencrypted" and should be reserved for cases where an attack is strongly suspected, cases where the error is unlikely to be simply the result of poor maintenance (outdated) or not wanting to deal with certificate authorities (self-signed).
Also note that the examples I gave are not necessarily the best. The true conditions should be determined by actual data. But, I sometimes see myself going to the http version of a (non sensitive) site to avoid the warning, that's retarded and browsers shouldn't encourage this behavior. Also, wanting to visit a broken https site once doesn't mean I want to add an exception forever.

Safari 9 on the Mac...

[Rick Zeman]

...has a menu option in the develop menu for "Treat SHA-1 Certificates as insecure." Nice having the flexibility to turn that on and off depending on need.